One of crucial components in visual tracking is the dynamical model of human motion, which is used for estimating the person's position in a sequence of images. Building a dynamic model of human motion is particularly difficult due to unpredictability of motion, ranging from Brownian-like to linear motions. We proposed a dynamic model composed of two components, first modeling abrupt motions and the other serving as a regularizer. We implemented the model in a recursive Bayes filter and showed that it allows more accurate and robust estimation of position as before.
COBISS.SI-ID: 7709524
We propose place/transition Petri nets (PNs) for the recognition and evaluation of complex multi-agent activities. The PNs were built automatically from the activity templates that are routinely used by experts to encode domain-specific knowledge. We extended the original PN formalism to handle the propagation of evidence using net tokens. The evaluation of the spatial and temporal properties of the actions was carried out using trajectory-based action detectors and probabilistic models of the action durations. The approach was evaluated using several examples of real basketball activities.
COBISS.SI-ID: 7389780
Problems with significant input-data uncertainty are very common in practical situations. In this paper we present a new approach to coping with data uncertainty, called the flexibility approach. The idea is to find a solution for each scenario such that, after a change in scenario, transforming from one solution to the other one is not expensive. We define two versions of flexibility and prove for both problems the NP-hardness as well as the non-approximability. We proceed in a standard way with the development of efficient algorithms for special cases of the general problem.
COBISS.SI-ID: 7087700
The moving least squares (MLS) is an essential approach that guarantees the locality of many meshless methods for the numerical solution of partial differential equations (PDE). The locality is implemented by hat-shaped MLS weight functions applied on local support domains. Three methods for the determination of the MLS support domain have been analyzed experimentally: constant radius, sampled and interpolated radius and continuously variable radius with selected numbers of support points.
COBISS.SI-ID: 23882023
We have investigated the approach based on a reduced number of ECG electrodes and reconstruction of the standard 12-lead ECG. The minimal number of electrodes has been determined, and their positions and distances optimized for each investigated person. Personalized transformation matrices have been obtained using multichannel ECG. It has been shown that a linear combination of only three independent potentials from wireless electrodes suffices for an accurate reconstruction, which could be advantageous for wireless sensor network (WSN) applications.
COBISS.SI-ID: 24198695